Thank you for your prompt response!
Sorry, this was more a C++ question. I'm still having a hard time to
disentangle all of these methods and languages.
It worked! I have included below the .R, .h, and .cpp files of the fx
from my previous e-mail. Just to help the next newbie ;) .
Thanks!
Marie
f1.cpp:
#include "f1.h"
SEXP f1(){
using namespace Rcpp ;
NumericVector x(1);
x = 9;
return x;
}
f1.h:
#ifndef _fxwithinfx_f1_H
#define _fxwithinfx_f1_H
#include <Rcpp.h>
extern "C" SEXP f1() ;
#endif
f2.cpp:
#include "f1.h"
#include "f2.h"
SEXP f2(){
using namespace Rcpp;
NumericVector x = f1();
x[0] = x[0] + 1;
return x;
}
f2.h:
#ifndef _fxwithinfx_f2_H
#define _fxwithinfx_f2_H
#include <Rcpp.h>
RcppExport SEXP f2() ;
#endif
f2.R:
f2 <- function(){
.Call( "f2", PACKAGE = "fxwithinfx" )
}
On 02/05/2012 4:18 PM, Dirk Eddelbuettel wrote:
Marie,
On 2 May 2012 at 15:49, Marie Auger-Methe wrote:
| Dear list,
|
| Still pretty new to Rcpp and C++. I am trying to make a package that has
| multiple Rcpp functions. Some of the Rcpp functions call other Rcpp
| functions and I am wondering what is the most efficient method to handle
| this situation. Here is a example of how I am handling the function
| within function (examplified using inline rather than .cpp and .h file):
Don't use inline for this. It generates randomized function 'names' at the
C++ level at it is tailored for use _from R_ and the C++ function pointer
gets hidden away in an R function.
It is time to think about packages. Inline is very convenient, but it has
occassional limits. You are hitting one now.
What you want is trivial and can be fixed the standard C / C++ by declaring
the function headers as eg in
extern "C" SEXP foo(SEXP a, SEXP b);
extern "C" SEXP bar(SEXP c, SEXP d);
before you implement the functions. Now you can call any way you want foo
calling foo, foo calling bar, bar calling foo, bar calling bar, ...
And of course you do not need to use the SEXP name(SEXP a, SEXP b)
convention use by R. You only need this when you call from R via .Call(). At
the C++ level you are open to any other (visible) C++ function -- which is
how we call other libraries.
My last piece of advice would be to write a mock C++ solution first. Write a
short main(), set up some data, call one function, have it call another.
Then work out Rcpp packages which you have meant to do for a while now :-)
and wrap it around your initial C++ program, replacing the main() function
which one you can call from R. And then you're done.
Dirk
_______________________________________________
Rcpp-devel mailing list
[email protected]
https://lists.r-forge.r-project.org/cgi-bin/mailman/listinfo/rcpp-devel
